The transcriptional response of third instar Drosophila larvae to food laced with caffeine, pyrethrum oil or neem oil was assessed using Agilent 4x44K printed microarrays.  Among the most notable responsive genes were those belonging to cytochrome P450 and Ecdysone-like kinase gene families.

Drosophila melanogaster from the genome reference line (y; cn bw sp Bloomington ID 2057) were raised on agar/semolina food and set to lay on grape juice laying plates with live yeast for 6 hours. Embryos were washed of yeast and placed on 2x agar following the protocol of Willoughby et al. 2006 (https://doi.org/10.1016/j.ibmb.2006.09.004). The resulting larvae were allowed to develop at 25C under constant light for 4.5 days and transferred to 2x agar laced with neme oil (5 mg/mL from Neemoil Australia Imports), pyrethrum oil (5mg/mL Botanical Resources Australia), caffeine (1.5mg/mL from Sigma) or onto untreated control plates for four hours. Larvae were snap frozen in liquid nitrogen, ground in a mixer-mill at 25kHz for 5 minutes and RNA extracted with TriSure (Bioline cat no. BIO-38032). Purified mRNA re-suspended in water were sent to the Australian Genome Research Facility who reverse transcribed into cDNA, labelled it with Cy3 or Cy5 and hybridized to Agilient 4x44k gene expression microarrays (018972 Oligo microarray version 1, 4x44k, G2519; https://doi.org/10.1016/S0076-6879(06)10002-6 ). Two arrays were used (Barcodes: 251897210115 and 251897210116), to provide a degree of technical replication. A pool of random RNA was prepared by mixing RNA from all four chemical stress conditions (Control, Neem, Caffeine and Pyrethrum). This pool was created to ensure that each transcript in each sample would have to compete with a differently dyed RNA transcript from the pool to hybridise with its microarray probe. This competitive hybridisation approach compensates for differing affinities between the probes and their RNA transcripts.

The Limma library (Smyth 2005 limma: Linear Models for Microarray Data: https://link.springer.com/chapter/10.1007/0-387-29362-0_23 ) was used in an R script to normalize the arrays.

The arrays were originally intended to be set up as technical replicates with the gasket order mirrored; the pool RNA in all gaskets labeled with Cy5 dye and the sample RNA labeled with Cy3. Due to a misunderstanding, the arrays were dye swapped and the gasket order randomised. This means that the gasket and dye effects for the Control and Pyrethrum groups are confounded. Initially the unwanted dye swap was thought to be a concern, but examination of the quality data for the arrays showed that there was very little dye effect for the non-confounded samples Caffeine and Neem. Therefore the unwanted dye swap does not appear to adversely affect the dye effect in the dye/array confounded Control or Pyrethrum groups either.